Yeah, I think it depends on whether you consider it a prediction if the recession occurs during the inversion (Harvey's thesis, IIRC, considers it a prediction if a recession occurs within 1-4 quarters after the inversion, regardless of whether the inversion is maintained into the recession itself. Plus, I think his approach was less predictive prior to the 1976+ time period, though that might have been a data noise artifact.)
If you accept Harvey's definition, then the major misses were the minor 10-2 inversion in the summer of 1998 (which prompted a quick round of Fed cuts that may have avoided a mild recession); the persistent inversion of 2000, which correctly predicted an economic slowdown but not a recession (in the US, anyway); the February 2006 inversion, which was in response to Fed attempts to cool down the housing market, which bit us a few years later; and the summer 2019 inversion, which I think was a true miss, though COVID pretty much upended any "normal" economic cycle, so it's possible that absent pandemic-era stimulus we might have had a mild recession. (Src: https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs and https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv)
As you say, it's an imperfect metric by any light, even though I buy into the less-strict traditional prediction model, so take it a bit more seriously when the lights start flashing. Regardless, for what my opinion's worth (which ain't much), I think this is looking more like '98 at best and '01 at worst, at least as long as the Fed doesn't overshoot and slam us into the tarmac.
If you accept Harvey's definition, then the major misses were the minor 10-2 inversion in the summer of 1998 (which prompted a quick round of Fed cuts that may have avoided a mild recession); the persistent inversion of 2000, which correctly predicted an economic slowdown but not a recession (in the US, anyway); the February 2006 inversion, which was in response to Fed attempts to cool down the housing market, which bit us a few years later; and the summer 2019 inversion, which I think was a true miss, though COVID pretty much upended any "normal" economic cycle, so it's possible that absent pandemic-era stimulus we might have had a mild recession. (Src: https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs and https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv)
As you say, it's an imperfect metric by any light, even though I buy into the less-strict traditional prediction model, so take it a bit more seriously when the lights start flashing. Regardless, for what my opinion's worth (which ain't much), I think this is looking more like '98 at best and '01 at worst, at least as long as the Fed doesn't overshoot and slam us into the tarmac.